研究開始時の研究の概要 |
This research will accelerate the revolution of designing complicated architected materials by removing guesswork from material design in a variety of applications. This work is based on deep learning with big data. A deep neural network will be trained using tens of thousands of structured data, which is similar to the way how species are differentiated and evolve by trial and error. The well-trained neural network is finally capable of generating flexible, tough auxetic metamaterials with extreme properties.
|
研究実績の概要 |
In this year, I have three main works, including a review article in terms of deep learning in mechanical metamaterials, a research article in terms of multiphase metamaterials with highly variable stiffness, and a research article in terms of text-to-microstructure generation using deep learning. In the review article, I provide a comprehensive overview of the capabilities of deep learning in property prediction, geometry generation, and inverse design of mechanical metamaterials. Additionally, I highlight the potential of leveraging deep learning to create universally applicable datasets, intelligently designed metamaterials, and material intelligence. In the second article, I propose three multiphase metamaterials derived from triply periodic minimal surfaces. The multiphase metamaterials possess highly variable stiffness based on thermally-induced phase transition. In the third article, I propose a new deep learning framework that can generate different and diverse material microstructures using text prompts. I have published 6 peer-reviewed papers on international journals and 1 patent during this academic year.
|